DSC World has to be one of the weirdest coded games out there, for sure. It uses a game engine similar to Crysis 1 in design where CPU clocks (of a single core) are vital for base performance, at about a rate of 1ghz = 25FPS (in simple areas) to 10FPS (flying over the most complex areas).
Here an example of how absurd it gets:
View: https://www.youtube.com/watch?v=pt0ImDL0Kys
We are talking about a CPU with twice the single-threading (and various order of magnitude in multi-threading), but it just sees a 25% FPS improvement across the run, which is much closer to the difference in clocks (5500ghz / 4800ghz) of 15%. Notice also the run uses a RTX 3080 Ti, which is stronger than the RX 7800 XT.
The devs instead reworking the engine, have just worked around it. You can notice how RAM can go easily up to 42GB in some peaks. RAM and RAM speed are key to squish out more performance. Intel 14th gen (14500 onward) is bottlenecked by RAM, too, so assume a 5-15% lower performance in DDR4 setups, 15% for... well, games like this, precisely, and 5% for more normal stuff. There is also absurd amounts of storage used to make up things, this game uses 300GB at times easily, lol, and is extremely picky with the SSD speed.
Now the CPU IPC can see more usage, during VR (which if they want, buy a Nvidia card then), if the game is modded and the mods uses a CPU logic nowhere near as weird, or in multiplayer, or in specific actions (weapon swaps/other game logic calculations peaks adjacent). Those are indeed affected by CPU's IPC, and the better it is, the fewer mhz are wasted in those tasks and can go into the main engine instead. This is also where the multi-core mode (barely) affects, too.
Anyway, went too long. Since the way the CPU is used is just too inefficient, nearly the totality of how graphics are processed and run is done locally by the GPU, with fairly little use requested from the CPU, so graphics mostly bottleneck at quality/resolution level. The RX 5700 XT can just do 45 FPS'ish average on High quality 1440p regardless the CPU, whereas there is players getting up to 120 FPS 4k with the RX 6800 XT (like here
https://www.reddit.com/r/hoggit/com...800xt_impressive_performance/?sort=confidence ).
Finally, if in doubt, there is plenty of tests of this game with the i9-9900K, sure, it is not the i5-9600K, but for the purpose of this game, both are very similar, 9900K has hyper-treading (unused by the game), has more L3 cache (just tangentially used by the game), has 8 cores instead 6 (unused by the game) and 400mhz more of stock clock (for a 8% uplift).
View: https://www.youtube.com/watch?v=dCzQjYSNZO4
The RTX 2060 is similar to his current GPU, and the RTX 3080 is similar to the RX 7800 XT. 50% uplift with a i7-9700K, and that is despite some RAM bottleneck going (both GPU and system RAM at 24GB), and graphics settings being put very modest low to medium, even then the i7-9700K is not bottlenecking it and with better graphics can easily get a 90% uplift with some tweaking.
Finally, if he wants to upgrade his CPU for stability/squish some extra FPS/mods/AI/multiplayer intense, a K model is definitively recommended, since clocks here are the bread and butter of the game, the lower clocked CPUs struggle. But if he only updates GPU now he will see significant results right away.
I personally recommend waiting to update the CPU since Intel K is going quite jam on W consumption in 12th-14th gen (my 13600KF can easily spike 250W), and Ultra 200 series (Arrow Lake) will go from 10nm++ to 2nm process, so it surely will be able to clock with better power efficiency and not punish the power supply much. He can also go Ryzen 7000 series right now, but clocks aren't way better than his current CPU, maybe 8000 series provide. DDR5 RAM will also provide another leap in performance, specially if installing more capacity, and the more the wait, the more the price drops for modules.